The Forever War
« previous post | next post »
Geoffrey Pullum, "Chomsky's Forever War", National Review 2/17/2022:
Few American linguists were puzzled when they saw the title of Randy Allen Harris’s book about events in their discipline between 1965 and 1975: “The Linguistics Wars.” Academic feuds are famously bitter, but the hostilities that Harris chronicled were unusual even by the standards of the humanities and social sciences.
Studying and interviewing linguists the way an anthropologist might study the culture of a belligerent primitive tribe, Harris produced his insightful but also entertaining book in 1993. Oxford University Press has recently published a revised and expanded edition (2021). Typically, historiography of arcane intradisciplinary wrangling among grammatical theorists would not get that far. The wide appeal of this book is probably due to its central character. Looming over the whole story is the charisma, influence, and personality of a single individual whose work utterly dominates the field: Noam Chomsky.
It is quite difficult to explain in nontechnical terms what triggered the linguistics wars, but let me try.
I'm not a syntactician or a philosopher. But based on my experience since 1965, when I first heard Noam Chomsky lecture on syntax, and 1972-75, when I was a graduate student at MIT, Pullum's account (of Harris's account) seems accurate to me.
Let me add one thing.
As Pullum notes, Chomsky is often associated with epistemological nativism, and has definitely had strong influences in that direction. Many brilliant researchers were thereby attracted to his cause(s), especially in the heyday of behaviorist empiricism. See e.g. Lila Gleitman et al., "The Impossibility of Language Acquisition (and How They Do It)", Annual Review of Linguistics 2019.
But in fact Noam's fundamental allegiance is to rationalism — he believes that the aspects of language that interest him don't have to be learned, not because they're somehow coded in the genome, but because they're a necessary consequence of the logic of the situation (maybe along with one little mutation giving rise to "recursion/merge" or whatever). For more on this, see "Chomsky testifies in Kansas", 5/5/2005, where I observe that (non-accidental) Darwinian evolution towards "innate" language abilities would be a kind of empiricism, just involving genomic learning rather than neuronal learning.
For one extended skirmish on the epistemological fringes of the Forever War, see "JP versus FHC+CHF versus PJ versus HCF", 8/25/2005.
For more Noam on the issues involved, see "Straw men and bee science", 6/4/2011 — perhaps especially what I wrote in response to a comment on that post:
In the very first lecture that I ever heard Chomsky give, in the fall of 1965, he covered several chalkboards with examples and rules dealing with "affix hopping", but he also mentioned in passing that to date, the only empirically adequate theory of human learning was Plato's notion that learning is just remembering things you experienced in previous lives. Hilary Putnam got up and protested "Surely you aren't really proposing reincarnation as a scientific hypothesis!", and Noam responded, "It's a better hypothesis than anything psychologists have come up with since then".
I'm not suggesting that Noam believed or believes in reincarnation, but it's clear he likes to shock people by setting Plato up as superior to the past 150 years of psychological research. In fact, he's just as opposed to the reconstruction of Plato in terms of the idea that learning takes place in the genome — see his arguments with Pinker and others.
Update — We also shouldn't forget that the start of the (last 60 years or so of the) "Linguistics Wars" was the campaign against linguistic "Structuralism", then the dominant paradigm, carried out vigorously by Noam and his acolytes. For an example of what that campaign was like as of the mid-1960s, see "Philosophical arguments about methodology", 5/7/2014.
Jerry Friedman said,
February 20, 2022 @ 10:55 am
Wow. Geoff Pullum's thesis in that article seems to be that Chomsky is a vicious liar who has never produced any useful academic work. Maybe the article is another shot in that war, which you imply is still going on.
[(myl) Geoff is definitely not a fan of Noam's, but I think that even in the NR piece you can see some respect for his real contributions. Certainly I don't endorse denying or minimizing Noam's contributions, though it's also relevant to recognize his role in over-politicizing approaches to linguistic analysis.]
I was curious about the political context, since the National Review is a conservative magazine and Chomsky is a famous leftist, as mentioned in the article. I now see why they liked the article.
[(myl) Indeed — and ironically, the associated politicizing of issues normally far outside NR's concerns is a shining example of what was wrongest about the events described in "The Linguistics Wars".]
Footnote to the title: The Forever War (1974) is a science fiction book by Joe Haldeman about a war with aliens that lasts over a thousand years. It won the Hugo, Nebula, and Locus awards. Haldeman is a veteran of the Vietnam War and taught creative writing at MIT, coincidentally, for many years.
[(myl) I should have mentioned that novel — thanks for bringing it up. From the Wikipedia article:
The novel is widely perceived to be a portrayal of the author's military service during the Vietnam War, and has been called an account of his war experiences written through a space opera filter. Other hints of the autobiographical nature of the work are the protagonist's surname, Mandella, which is a near-anagram of the author's surname; Mandella being a physics student, like Haldeman, as well as the name of the lead female character, Marygay Potter, which is nearly identical to Haldeman's wife's maiden name. If one accepts this reading of the book, the alienation experienced by the soldiers on returning to Earth – here caused by the time dilation effect – becomes a clear metaphor for the reception given to US troops returning to America from Vietnam, including the way in which the war ultimately proves useless and its result meaningless. He also subverts typical space opera clichés (such as the heroic soldier influencing battles through individual acts) and "demonstrates how absurd many of the old clichés look to someone who had seen real combat duty".
I returned in 1970 from a tour in Vietnam, and so when I read Haldeman's book, those resonances (about the effects of coming back to what we called "the world") were personal. I started grad school shortly after getting out of the Army in 1972, and found the on-going Linguistics Wars similarly evocative and alienating. This reaction was a large part of my decision to forego a couple of academic job offers in favor of working in an industrial research laboratory for the next 15 years.]
David Marjanović said,
February 20, 2022 @ 12:13 pm
The latest shot in the linguistics wars: an open-access paper on how language can indeed be learned from input alone.
[(myl) This is part of a more general migration back towards empiricism, largely (but not entirely) fueled by "deep learning". But notice that it's entirely consistent with Noam's current (?) "only merge" position, i.e. that the only relevant innovation in human language (in his "narrow" sense) is "recursion", in his definition as the ability to create complex structures via hierarchical combination of simpler ones.]
DJL said,
February 20, 2022 @ 1:27 pm
@ David Marjanović
Worth noting that that paper is about extrapolating formal grammars from strings of elements, and thus, very possibly of rather limited relevance to the study of the acquisition of natural language, as has been noted ad nauseam over the years (it constitutes a rather narrow and technical point, and perhaps an interesting one, but not one that directly relates to the study of natural language).
The stuff about 'deep learning' is a bit of a red herring – not a psychological mechanism, and the resultant input-output correlations bear little resemblance to natural language.
By the way, whatever happened to Pullum on this blog?
Brett said,
February 20, 2022 @ 3:51 pm
While the substance of this post was interesting, I was most struck by an issue with its pragmatics: It refers to Geoffrey Pullum (whom I suspect that Mark Liberman would consider a personal friend) by his last name but Noam Chomsky by his first name. Referring to Noam Chomsky somewhat pejoratively as just "Noam" is something that I have been exposed to for essentially my entire life, mostly via my father, who was also a MIT graduate student in the first half of the 1970s (although in applied mathematics, not linguistics)
[(myl) When I was a student taking courses from him, everyone referred to "Noam Chomsky" as "Noam" — nothing pejorative about it. And the practice continues among those who know him. And so it would seem odd to me to refer to him repeatedly as "Noam Chomsky" or "Prof. Chomsky" or even "Chomsky" in a blog post, though I've probably used all of those at one time or another — and I did refer to him as "Chomsky" the first time in the post above. Geoff once upbraided me for not using his last name in a blog post reference, on the grounds that casual readers won't know who "Geoff" is, so I guess that pushes me in the direction of "Pullum" as the right referring expression.
Philip Taylor said,
February 20, 2022 @ 4:54 pm
A comment and a question. Comment: I don't find referring to Noam Chomsky as just "Noam" in the least pejorative; rather endearing, really.
Question: "a shining example of what was wrongest about the events described" — "wrongest" ?! Humorous, personal, topolectal, or what ?
bks said,
February 20, 2022 @ 5:26 pm
On the basis that aesthetics is the opposite of anesthetics, Chomsky was undoubtedly the greatest academic artist of post-WWII America. When I got to college in 1969 there was a cross-disciplinary maelstrom surrounding Chomsky involving Psychology, Anthropology, Philosophy, Political Science and, of course, Linguistics. I don't think there's been anything like it since.
David Marjanović said,
February 20, 2022 @ 6:36 pm
Yes; if every sentence of three words contains two instances of MERGE, then the paper doesn't falsify that; I don't think it's falsifiable at all.
What the paper utterly trounces, burns down, and then trounces the ashes of to aerosol particles is "the poverty of the stimulus", the basic, foundational, crucial assumption of everything Chomsky wrote about linguistics before Minimalism.
Then why was the study made at all? I think the paper* explains its relevance pretty well.
* PNAS is a glamour mag. What I linked to is called the paper, but it's just an extended abstract. The actual paper is called "Supplementary Information" ("SI").
Here's a system that cannot have any innate language-specific abilities, and yet it learns a wide range of grammars. That was supposed to be impossible, and explicitly stated to be impossible again and again for decades just because it seemed self-evident to Chomsky and a few other people.
Would you prefer "most wrong"?
Or does "wrong" feel like part of a binary to you, so it doesn't come in degrees? I couldn't agree with that; "what was wrongest" is that which contained the greatest amount of wrongness, and that's a continuous variable.
[(myl) And of course, some ideas are "not even wrong"…]
AntC said,
February 20, 2022 @ 9:26 pm
"the poverty of the stimulus"
Chomsky's granddaughter's kitten hasn't/won't learn to play board games, etc — although some cats can play the piano, allegedly.
How may genes/mutations do we have to suppose the kitten lacks, to explain all these inabilities?
Victor Mair said,
February 20, 2022 @ 9:45 pm
I bought The Linguistics Wars almost as soon as it came out in 1993, because I wondered how anything as cerebral as linguistics could lead to war. After I read it, I was disappointed because it was hard for me to tell exactly what they were fighting over. In the long run, however, the book proved successful, because it alerted members of the public to the fact that there was an academic discipline called "linguistics", and that its practitioners took their work seriously.
Seth said,
February 20, 2022 @ 10:31 pm
@ Victor Mair It's not strange that people who devote their lives to studying a topic, who are among the world's experts on that topic (or want to be), care very deeply and emotionally about what is correct in the area of that topic. There's something at work like the Mel Brooks quote "Tragedy is when I cut my finger. Comedy is when you fall into an open sewer and die.". Roughly "Grave insult is when you cut me off traffic. Triviality is if I say your entire life's work is wrong". There's also enough of a connection to broad nature vs nurture philosophical issues that can make for contention.
Julian said,
February 20, 2022 @ 11:31 pm
How does a child learn language while a kitten can't?
Maybe because a human being is just generally smarter than a cat?
[Apologies if this comment seems naive to people who actually know something about this topic]
DJL said,
February 21, 2022 @ 4:21 am
@ David Marjanović
The study of formal language learning is certainly worth conducting in its own right, and some of that stuff is relevant to linguistics too, possibly as a lower bound all grammatical theories must attain in terms of expressive power (ie, your theory must generate a language that is in the right class – is it mildly context sensitive still?).
My point, rather, was that natural languages are not like formal languages, natural language grammars are not like formal grammars, and thus any proof from formal language learning regarding what formal grammars can be extrapolated from the set of strings the system is exposed to (the positive data) is a rather narrow point, and not, ipso facto, a proof regarding natural language language acquisition. And, of course, no deep learning model has actually learned any natural language grammar, so the point really is moot.
Despite claims to the contrary, Chomsky has actually been very clear about the limited impact of formal language theory on the study of language acquisition for decades; what he says about natural language acquisition, and the poverty of the stimulus, does not depend upon any of the stuff from formal language theory.
Oh, and I know what PNAS is, and what a paper is, and what an abstract, even what SI is (which is never the actual "title" of a paper), and I have even read the thing, so I don't need the silly patronising.
SusanC said,
February 21, 2022 @ 5:33 am
A cat will almost certainly attempt to walk on your piano, unless prevented somehow. Cats are like that. But cats are probably not motivated by the sound it produces.
This now has me wondering if a cat toy that makes different musical notes would get played with; and more generally, if cats have any understanding of music.
Philip Taylor said,
February 21, 2022 @ 6:35 am
David M — yes, "most wrong", in contexts in which "wrong" is not absolute. "wronger" and "wrongest", while attested in the 18th and 19th centuries, have been replaced by "more wrong" and "most wrong" in more recent times. And while "more wrong" is relatively common, "wronger", "most wrong" and "wrongest" are noise-level by comparison.
J.W. Brewer said,
February 21, 2022 @ 10:12 am
I've never personally referred to Chomsky mononymously as "Noam," but it makes perfectly good pragmatic sense to do so, because for the relevant generational cohorts it was a vanishingly rare given name in the U.S., so the possibility of ambiguity was quite low, whereas while "Geoff[rey]" was still pretty rare in the U.S. in the year Chomsky was born (1928) but it was on the rise by the time Pullum was born (1945, not in the U.S.), and from 1947 through 1991 was one of the 300 most common names given to US-born baby boys (peaking in rank at #167 in 1975) before falling out of favor in the new millennium. It was always (in the U.S. at least) a much-rarer variant of "Jeff[rey]," which had a somewhat similarly-shaped rise and fall arc but peaked in the top 10 for a few years including my own year of birth.
So the median American my age or myl's age has probably known multiple Geoffs (although even more Jeffs) but well may not have known any Noams or even heard of any other than maybe Chomsky. I have known in person (as a one-time work colleague) exactly one Noam, and I was past 45 when I first met him.
And yet and yet … due to the cyclicality and churn in naming practices things have finally turned around in the new millennium, and the Social Security Administration's statistics for 2020 show only 46 newborn Geoffreys compared to 98 newborn Noams. So when today's toddlers grow up to be feuding academics, the pragmatics of how to refer to them may have changed.
Carl Voss said,
February 21, 2022 @ 3:56 pm
Philip Taylor, I find "wrongest" superior to the proposed "most wrong" in Liberman's sentence. Of course, we're American.
Jerry Packard said,
February 21, 2022 @ 4:11 pm
"We also shouldn't forget that the start of the (last 60 years or so of the) "Linguistics Wars" was the campaign against linguistic "Structuralism", then the dominant paradigm, carried out vigorously by Noam and his acolytes."
It is quite ironic that all of Chomsky and acolytes' work during the 'campaign against linguistic Structuralism' explicitly dealt with the generation and manipulation of structures, despite the theory's cognitive and anti-behaviorist bent.
Also, the statement "Plato's notion that learning is just remembering things you experienced in previous lives." seems much more reasonable if we delete the pronoun 'you', implying that the 'remembering' is of things experienced not by us, but rather experienced by those that have preceded us.
Philip Anderson said,
February 21, 2022 @ 5:23 pm
@Carl Voss
Maybe there is a US/UK aspect, since *wrongest is just wrong for me as well (and for autocorrect).
Victor Mair said,
February 21, 2022 @ 6:16 pm
When I read "wrongest", it struck me as jarring, but then I immediately took it as Mark being ironic, and it became all right.
Lester said,
February 21, 2022 @ 10:06 pm
Echoing DJL: I'm delighted to see Prof. Pullum appear again (even if indirectly) in this blog. I sorely miss the brilliant insights and piquant writing he used to provide.
Bathrobe said,
February 21, 2022 @ 10:52 pm
In the Lila Gleitman paper (The Impossibility of Language Acquisition (and How They Do It)) there is an experiment concerning the verb 'to think' which looked at situations of false belief and true belief. Interestingly, Chinese uses separate verbs for false belief (以为 yǐwéi) and for the expression of what the speaker actually believes to be true (想 xiǎng).
Bathrobe said,
February 21, 2022 @ 10:55 pm
And of course there is 认为 rènwei, which means 'take the view that', which is less speculative (if you will) than 想 xiǎng.
Peter Grubtal said,
February 22, 2022 @ 2:48 am
When I first became aware of Chomsky and his ideas, it was the famous contrast "John is eager/easy to please" which looked to me at the time like a key insight. Later on, I began to wonder whether the "easy to please" bit isn't perhaps a unique characteristic of English, which is expressed quite differently in the fairly closely related languages French and German, for example.
Another peculiarity of English which struck me was "I was given a piece cake", which once again, in German would be expressed much more logically with a dative, and in French with a different grammatical construction.
So 50 years on has there been any progress in elucidating the deep universal structure underlying these contrasts, or are they just vagaries of English?
Seth said,
February 22, 2022 @ 3:09 am
@ Peter Grubtal – my very outside-observer thought is that whatever one says about the past 50 years, the next 50 years are going to be very kind to Chomsky's views given the advances from computers and data.
Philip Taylor said,
February 22, 2022 @ 5:41 am
Peter G — « Another peculiarity of English which struck me was "I was given a piece cake" » — Could you clarify, please ? Is there a missing "of" before "cake" or is this an Americanism on the same lines as "a couple days" where in <Br.E> an "of" would be required ?
David Marjanović said,
February 22, 2022 @ 6:23 am
Even on a blog like this, most readers don't have enough experience with academic publishing to know all of these things, so I thought I was making a reasonable assumption. Sorry. (I don't understand why you bring up titles, though.)
I do find it strange that you don't seem to find it worthwhile to explain your point. So, "Chomsky has actually been very clear about the limited impact of formal language theory on the study of language acquisition for decades" – why does the kind of formal grammar used in the study have so little impact on the study of language acquisition in Chomsky's view? You're only telling us that Chomsky says so.
(I'm aware of the claim that Chomsky denies the very possibility of learning through analogy. That's so strange I have trouble believing it's a fair description of Chomsky. But if it is, then of course the new paper has no impact on his concept of the poverty of the stimulus…)
AntC said,
February 22, 2022 @ 6:29 am
the famous contrast "John is eager/easy to please" which looked to me at the time like a key insight.
Those two sentences previously on Language Log. See a comment from a 'Philip' and its correction by a 'Chad'. And see several pointers from there on 'diagramming sentences'.
TL;DR " grammar-school children of the 19th century learned more about linguistic analysis than most graduate students in English departments do today. "
(I learnt more about English syntax in Latin classes than in English. The Latin master made us re-write the English into a stilted form with tags and bracketing.)
The more interesting History-of-ideas question is why did Chomsky get credited with 'discovering' (or at least publicising) something that was already (or perhaps I should say 'used to be') commonplace? The same q applies for his poaching from "Categorial Grammar", as Pullum mentions. I already knew of it from a) Logic (SKI Combinators), b) Functional Programming in Lisp, and c) a Linguistics textbook by David Crystal.
David Marjanović said,
February 22, 2022 @ 6:29 am
I bet it is; I've never encountered "a piece cake" in any kind of English. I'm sure the point is that, e.g. in German, I'm not what's given in "I was given a piece of cake" – the cake is, and it was given to me, so I go into the dative, which has generally merged with the accusative in English, so by German rules you'd expect me instead of I.
I wonder if *(to) me was given a piece of cake got crossed over with I got a piece of cake.
Peter Grubtal said,
February 22, 2022 @ 6:53 am
Philip, David –
yes, in my 50 years of proofreading, there has been some progress, but as you point out there's still room for improvement.
I'd noticed the mistake before you posted and thought it would be allowed to pass, but shame on me: I owe it to the readers to post the correction myself.
DJL said,
February 22, 2022 @ 7:02 am
@ David Marjanović
But I HAVE made my point, and it wasn't a matter of simply stating what Chomsky says (such a silly thing to say, in any case).
To repeat: natural languages are not like formal languages (a distinction between structures and strings of elements, say), and the grammars linguists propose to account for natural languages are not like the formal grammars of formal language theory, the latter of which are symbol rewriting systems, effectively. And, thus, a proof that some formal grammars can be extrapolated from strings of symbols, which is what the paper you linked to is about, is not a proof that is directly related to the acquisition of natural language.
I should stress that I am not defending Chomsky per se here; I'm simply stating that the paper you linked to doesn't really disprove any of the repercussions or issues that arise out of so-called poverty-of-the-stimulus arguments (and I was also pointing out that it is nonsense to say that any deep learning model has learned any natural language).
As for the mention of 'title', well, that is clear too, since you stated that 'the actual paper is called "Supplementary Information" ("SI")', which is obviously a mistake – that's not the name of the paper (its title).
JPL said,
February 23, 2022 @ 3:26 am
David Marjanovic links to a paper (one of his many) by the philosophically aware Esa Itkonen, and this, together with Geoff's mention of (the late) PAM Seuren, whose writings, such as Semantic Syntax, embody the essence of what seemed to me was the central question dividing the combatants in the linguistics wars, points to the reason Chomsky's theoretical approaches will probably not lead to a full understanding of language as a human phenomenon: his avoidance of the problem of meaning. Why has Chomsky been so hostile to the idea of taking meaning as an object of study? (I.e., as a puzzle to be figured out; I don't mean as used in the role of evidential support) It seems like, in spite of his refutation of behaviourism, he was not able to fully free himself from the conventional strictures of Machian phenomenalism endemic in those early days. The roots of his approach to language seem to have been in the tradition of formal language practice, such as Hilbertian formalism and the Vienna Circle in its more "positivistic" tendencies, in particular the new ideas of Carnap. (Take a look at The Logical Syntax of Language and Meaning and Necessity and you will feel a sense of intimations of prefiguration.)
JPL said,
February 23, 2022 @ 4:04 am
I mean, if one is going to try (and Chomsky was not) to achieve the aim of the other aspect of the Vienna Circle (see J. Alberto Coffa, The Semantic Tradition from Kant to Carnap) and try to flesh out the Kantian conception of the role of the (epistemic) subject's instruments of knowing in judging the adequacy, effectiveness, accuracy of the expressions of attempted understanding of the empirical world in scientific discourse, then one can not ignore the phenomenon of expressed meaning in the use of human language, and how expressed meaning relates to, is about, the world as an intentional object. (I know Chomsky's philosophical discussions are idiosyncratic, but why does he never talk about Kant?) If Chomsky regards language primarily as an instrument for the expression of thought, which he has said he does, and which I would say is correct (and so far we have not really gone beyond Frege's view), then you have to go on to say, "the expression of thought about the world". So what about the thought? Where is Chomsky's account of the thought expressed, as opposed to the expression of the thought? (These are two distinct intentional objects; Chomsky (and he is not alone in this) does a lot of what Hilary Putnam said is what mathematicians tend to do in their practice of talking about their notations: "point to the inscription and think the meaning". (What's going on in what he calls "bare syntax"?)) How does the Chomskyan dogmatic conception of the autonymy of (morpho-)syntax and the computation wrt perceivable forms connect with the world? He has abjured the notion of reference, but meaning expressed in language use must connect with the world somehow or other. I mean, again, if you take a point of view of Deweyan pragmatics (as the critique of action) on the human use of language, people are doing more than just making noises; they are purposefully trying to mean something by those noises, and something rather specific. How is that "something meant" possible? Also, Chomsky seems to have held the mistaken conception that the empirical scientific pursuit of the problem of the meaning expressed in human language use can only be done by the field of psychology, which he said has never said anything useful about it and probably never will, so you might as well not bother with it. But I think that it is a mistake to think that a scientific study of meaning as an empirical object necessarily has to be psychological; something else is going on there. In fact, I could be wrong, but I think that both Itkonen and Seuren were opposed to the "psychologistic" approach to the study of meaning, as well as to the reliance on non-empirical formal language practices, such as model theory. (I'm sorry, I knew I shouldn't have started on this. It's the not even the tip of the iceberg.)
BTW, I'm curious about what the new material is that is included in this new "expanded edition" of The Linguistics Wars.
Philip Taylor said,
February 23, 2022 @ 4:18 am
JPL — "I'm curious about what the new material is that is included in this new "expanded edition" of The Linguistics Wars" — I'll try to (remember to) let you know when my Amazon copy arrives. It is currently marked "Release date: Thursday, 28 April 2022" with an estimated delivery date of 03-May-2022 .
Alexander Clark said,
February 23, 2022 @ 4:24 am
DJL says "To repeat: natural languages are not like formal languages (a distinction between structures and strings of elements, say), and the grammars linguists propose to account for natural languages are not like the formal grammars of formal language theory, the latter of which are symbol rewriting systems, effectively."
These are reasonable points, but it's worth noting that lots of formal language theory is concerned with other systems that aren't string rewriting systems (e.g. automata, MSO logic and finite model theory, or Smullyan's elementary formal systems ), and that there is a vast literature on what are called tree grammars, where what are generated are trees not strings, and finally that Stabler's minimalist grammars are weakly and strongly equivalent to Multiple Context-Free Grammars which are a mild generalisation of the familiar context-free grammars. So while I completely agree with the limitations of formal language theory as it was say 50 years ago, the continued dismissal of it by Chomsky and others is a little puzzling.
randy harris said,
February 23, 2022 @ 1:51 pm
Victor Mair said:
I bought The Linguistics Wars almost as soon as it came out in 1993, because I wondered how anything as cerebral as linguistics could lead to war. After I read it, I was disappointed because it was hard for me to tell exactly what they were fighting over.
I’m sorry that I may not have been as clear as I should have been about what they were fighting over. Maybe I was too far into the trees to see the forest; perhaps the new edition suffers from this as well. But there is a pretty brief answer I can give: first, meaning in a (Chomskyan) transformational grammar; in the end, everything. In the beginning they were fighting over a very specific question, whether transformations altered meaning in the context of an Aspects theory of language, approaching it largely as an empirical question: does surface structure x have a different meaning than deep structure y as the result of transformation z? But that spiralled quite rapidly into all sorts of methodological questions, scope questions, data questions, architecture-of-theory questions, the relevance of psychology, sociology, philosophy of language, logic, and so on, until they were fighting over almost every square centimetre of language and linguistics (almost; phonology played a very small part). There were really very, very few points of contact between principals like Lakoff and Ross on one side, Chomsky and Jackendoff on the other. But there was also a flowering of frameworks as the battles ended occupying most points between them.
JPL said:
BTW, I'm curious about what the new material is that is included in this new "expanded edition" of The Linguistics Wars.
I cut out all of the pre-Syntactic Structures material at one end, which required a new approach to Chomsky’s rise, including a somewhat bigger role for machine translation and computer science. At the other end, I follow the ‘greening of linguistics’ trajectory out of the period through Cognitive Linguistics, Construction Grammar, and Frame Semantics, and the ‘right of salvage’ trajectory through Principles and Parameters, Minimalism, and Biolinguistics. In the core battle chapters, I’ve refined some of my understanding (and explication) of the developments (for instance, correcting my insufficient account of Robin Lakoff's role, and showing that GSers were relying on metonymy for counter-examples without apparently realizing there was such a thing as metonymy); I've also ). As a rough estimate, I’d say it’s probably 40% new stuff.
randy harris said,
February 23, 2022 @ 1:53 pm
(Oh, and about the publication date(s). The US date was in October 2021, so there are copies in circulation. The UK/global date was scheduled for January, but for supply-chain reasons has been pushed now to 28 April.)
DJL said,
February 24, 2022 @ 6:54 am
@ Alexander Clark
Agreed, and thanks for the comment. I know that literature well, which I have used in my own work (especially in a book on recursion I published 5 or so years ago), but I do think the point I was making is true concerning the PNAS paper linked to above (and the related comments about POS arguments, deep learning, etc.).